Introduction

Interviewing suspects is a crucial component of police investigations, often yielding statements that contribute to both just and wrongful convictions. Ideally, suspect interviews yield accurate information. Studies based on investigated self-reports existing studies often focused on the suspect’s experience during the interview (e.g. Cleary and Bull 2019), and observational methods (e.g. Feld 2006; Leo 1996;  Tersago et al. 2020) have shed light on what happens in the interview room. Only a few studies, however, have examined the perspective of police officers. Kassin et al. (2007) used a self-report survey to question North American police officers about their use of different interviewing techniques, the duration of interviews, (false) confession rates, and electronic recording. More recently, a European project (coordinated by Maastricht UniversityFootnote 1 in collaboration with John Jay College of Criminal JusticeFootnote 2) was organized to fill this gap. Following this project’s earlier publications on suspect interviewing in the Netherlands (Vanderhallen et al. 2022) and Spain (Schell-Leugers et al. 2023), the present research focuses on the beliefs and practices of police investigators in Belgium.

Approaches to Suspect Interviewing

Meissner et al. (2014) broadly distinguished between accusatorial and information-gathering approaches to suspect interviewing. The accusatorial model is guilt-presumptive, confrontational, and designed to obtain confessions (for a critique, see Kassin 2006) and is commonly used in the United States, Canada, and many Asian countries (Costanzo and Redlich 2010; Kelly et al. 2013; Leo 2008). One of the accusatorial approaches is the Reid technique (Inbau et al. 2013), which is a nine-step process in which the police (1) isolate the suspect; (2) confront that suspect with accusations made with certainty, often bolstered by the presentation of evidence (even if that evidence is false), and refusing to accept denials; and (3) offer sympathy and minimization themes that downplaying the moral seriousness of the crime, a tactic that implies that confessing will result in more lenient treatment (for a critique of the Reid technique, see Kassin 2006). As a result of this psychologically tactical, guilt-presumptive process, and its focus on a confession, research has shown that accusatorial techniques increase the risk that innocent suspects might falsely confess (Kassin et al. 2010; for a meta-analysis, see Meissner et al. 2014).

In contrast, the information-gathering approach aims at collecting accurate information rather than necessarily obtaining a confession. Instead of presuming a suspect’s guilt, the approach centers around gaining the suspect’s cooperation, communicating with the suspect with decency and empathy, asking open-ended questions, exhibiting a genuine interest in the suspect’s account, and confronting the suspect with evidence or contradictions in their account in a more neutral way (Malsch and de Boer 2019). The information-gathering style is used in the PEACE model of interviewing (Milne and Bull 1999; Walsh et al. 2017).

Suspect Interviews in Belgium

Suspect interviewing practice in Belgium has changed considerably over the past 25 years. In fact, solidly structured legislation on suspect interviewing and suspect rights was only introduced in 1998 (“Wet Franchimont”, Sv.), but this legislation did not precisely dictate any particular interviewing methods (Malsch and de Boer 2019). In the absence of uniform Belgian legislation on suspect interviewing, police services long followed North American procedures for suspect interviewing, incorporating both accusatorial and information-gathering interviewing techniques into their own police training programs (Vanderhallen et al. 2017). Considering accumulating research on the risk of miscarriages of justice posed by accusatorial methods, standard Belgian interviewing practice has shifted to a more neutral, information-gathering approach (Vanderhallen and Vervaeke 2014; Vanderhallen et al. 2017; Van de Plas 2007; Verhoeven 2018; Walsh et al. 2017). The Basisverhoortechniek [Basic Interview Technique] that presently governs standard suspect interviewing procedure in (all of) Belgium strongly reflects this shift as it centers around obtaining correct information and dictates neutral, non-suggestive communication with suspects (Van de Plas 2007).

Under the Salduz Act, suspects in Belgium have several legal rights, of which they must be properly informed before the interview commences (Art. 47bis §2 Sv.). Among the most notable of these rights are the right to remain silent and the right to legal assistance, which were only legally implemented as caution with the Salduz Act in 2012. The right to silence is based on the right against self-incrimination (nemo tenetur) and allows suspects to choose to not answer (some) questions beyond providing their identity to interviewers. The right to legal assistance holds that suspects have the right to consult with a lawyer, either in the form of a confidential consultation prior to or assistance during the police interview. During the interview, a lawyer may ensure that police do not treat the suspect unlawfully. Since the implementation of Salduz Bis in 2016, lawyers may also ask for clarification and request specific investigative acts to be performed (Malsch and de Boer 2019).

Making written records of suspect interviews is part of standard Belgian interviewing procedure (Malsch and de Boer 2019), but making electronic recordings of interviews is not as it is not mandatory (Art. 112ter, Sv.). Although the Salduz Act in 2012 did obligate police investigators to record suspect interviews, since Salduz Bis, it has become a mere recommendation considering the difficulties of funding recording equipment and storing recordings (Maegherman and Vanderhallen 2018). Because suspect interviews are not electronically recorded by default, gaining insight into what goes on in Belgian interview rooms is difficult.

Police Officers’ Beliefs and Practices

The development of interview manuals from accusatorial to information-gathering interview methods does not automatically imply that police officers’ beliefs and practices align with this shift. This issue has been researched in the North American context (Kassin et al. 2007) and in some European countries (e.g. the Netherlands (Vanderhallen et al. 2022) and Spain (Schell-Leugers et al. 2023)), but insight in Belgian police officers’ beliefs and practices about suspect interviewing has remained incomplete.

Existing research, though limited and outdated, suggests that only 21.9% of Belgian police considered obtaining a truthful statement the main aim of a suspect interview (Ponsaers et al. 2001). Despite the subsequent development of the Basisverhoortechniek, which emphasizes this objective (Vanderhallen 2007), its current adoption among Belgian police remains uncertain. Additionally, a minority of officers continue to employ accusatorial tactics, such as presenting impermissible evidence and expressing certainty of guilt to suspects (Ponsaers et al. 2001). Recent studies suggest a gap between officers’ reported support for information-gathering methods and their inconsistent application in practice (Beazley et al. 2021b).

Furthermore, police officers often try to circumvent suspects’ rights or persuade suspects to waive them. Despite early evidence of suspects’ awareness of their rights after the Salduz Act implementation (Penne et al. 2013), recent findings indicate inconsistent efforts by interviewers to confirm suspects’ awareness of their rights (Beazley et al. 2021a, b). Although only a small minority of suspects in Belgium uses the right to remain silent fully (Beazley et al. 2021b; Penne et al. 2013; Vanderhallen et al. 2014), a minority of Belgian police officers appear to not respect and accept a suspect’s decision to use this right (Beazley et al. 2021b). Instead, they attempt to pressure the suspect into waiving this right using minimization and maximization techniques. Belgian police officers further appear to believe that suspects who use their rights are more likely to be guilty than suspects who waive their rights (Beazley et al. 2021b).

Perceptions of guilt can also influence interviewer behavior, with research indicating police tend to overestimate their ability to detect deception (Kassin et al. 2003, 2007). Even after specialized training, police officers fail to outperform laypersons despite being more confident about their judgments (DePaulo and Pfeifer 1986; Elaad 2003; Kassin et al. 2003, 2005; 2007; Meissner and Kassin 2002). The actual accuracy of professionals’ deception detection judgments varies from 45 to 60% and averages only just above chance level (54%; Bond and DePaulo 2006; Vrij 2000). Inaccurate judgments can have grave consequences, especially if police investigators presume a suspect is guilty and resort to confession-focused interviewing methods (Leo and Drizin 2010).

The Present Research

Across two studies, the present research investigates Belgian police officers’ beliefs and practices regarding suspect interviews. Using an online self-report survey adapted from Kassin and colleagues (2007), officers were questioned about the goal, duration, frequency, and recording of suspect interviews; suspects’ use of legal rights; their ability to distinguish between truth and deception; (false) confession rates; and techniques used to interview suspects. In Study 1, Flemish police officers share their beliefs and perceptions, whereas Study 2 covers the beliefs and perceptions of officers from the French- and German-speaking parts of Belgium.Footnote 3 Ethical approval for the European survey (including the two studies presented here) was granted by the Ethics Committee of Maastricht University.

Study 1

Method

Participants

A total of 188 Flemish police officers agreed to participate in the study. However, 49 responses were omitted because fewer than 24 out of the 27 interviewing technique questions had been answered. Data distributions were checked for normality and outliers were omitted.Footnote 4 Checking for exclusion criteria resulted in the removal of an additional nine respondents, yielding a final sample of 130 respondents.

Materials

Police officers were asked about their beliefs about suspect interviews using an adapted version of Kassin and colleagues’ (2007) self-report questionnaire. To tailor the original, North America-oriented questionnaire to the Belgian legal context, several items were replaced. For example, questions about Miranda rights were replaced by questions about the right to legal assistance and the right to silence, and interviewing techniques were evaluated based on their relevancy for the European Member States. The adapted questionnaire was translated into Dutch following Pan and De La Puente’s (2005) instructions for translating data collection instruments. The final agreed-upon version was pretested in a semi-structured interview pilot with 15 Flemish police officers and adjusted according to their feedback.

Part I of the questionnaire inquired about participants’ age, gender, years of experience working in law enforcement, the number of interviews they had conducted over the past 12 months, the type of cases they mainly investigated, and whether they worked for the local or the federal police. Part II contained questions about respondents’ beliefs and experiences regarding suspect interviews. Officers were asked if they typically record suspect interviews (yes, always/yes, in specific circumstances/no) and if so, how (audio/video/sometimes audio, sometimes video), and if they believed interviews should be recorded (yes, audio/yes, video/no). Officers were also asked about the average duration and frequency of suspect interviews, the duration of the longest interview they had ever conducted, and the aim of suspect interviews (to obtain a confession/to obtain the truth/to gather accurate information). They were further asked to estimate their ability to distinguish between truthful and deceptive statements and whether this ability differed depending on whether the suspect was confessing or denying (no, it does not make a difference/yes, I am better at judging whether a confessing suspect is telling the truth or lying/yes, I am better at judging whether a denying suspect is telling the truth or lying). Finally, respondents were asked to estimate (false) confession rates. In Part III, officers rated on a 5-point Likert scale (1 = never; 2 = on rare occasion; 3 = sometimes; 4 = often; and 5 = always) how often they used each of 27 suspect interviewing techniques.

Procedure

Participants were recruited through the Belgian Police Academy. Police officers who had participated in a course on interviewing within their training or an advanced interview training in the past ten years at the National Criminal Investigation Academy (“Nationale Rechercheschool”) in Flanders were contacted via email. The email contained an invitation to take part in the study, information about the aim and anonymous nature, the estimated duration, and a hyperlink to the questionnaire. Upon opening the survey, respondents received an informed consent form and only received the questionnaire upon providing consent.

Results

Of the 130 Flemish police investigators who responded to the survey, 115 (88.5%) were employed by the local police, while 15 (11.2%) worked for the federal police. Overall, 92 males (70.8%) and 38 females (29.2%) participated in the study. On average, respondents were 41.68 years old (SD = 7.93), had been employed by the police force for an average of 19 years (SD = 8.44). Respondents came from various departments (see Table 1), but most indicated they work at the algemene recherche [general criminal investigation], dealing with various types of crime.

On average, respondents had 9.8 years of experience conducting suspect interviews (SD = 6.55) and indicated having conducted 42.75 suspect interviews in the last year (SD = 39.02, N = 130). Regarding their own interviewing skills, respondents evaluated themselves with an average of 7.34 (SD = 0.85) on a scale from 1 (not good at all) to 10 (excellent). The majority (85%) of investigators reported that they had received training on how to conduct suspect interviews.

Table 1 Distribution of departments represented in the sample
Full size table

Electronic Recording

When asked whether they typically electronically record suspect interviews, 6% stated that they always record interviews while 40% reported that they do so in specific circumstances and 54% indicated that they do not record suspect interviews. Of the 60 officers who used electronic recordings, the most common method was by videotape (88.30%) with the remaining 11.67% stating that they sometimes use audiotape and sometimes videotape. When asked about their opinion on whether suspect interviews should be recorded, 78.46% said yes. Within this group, 85.29% favored videotaping and 14.71% favored audiotaping.

Suspects’ Rights

Respondents were asked to estimate, based on their experience, how many suspects use their right to legal assistance before and during the interview as well as the right to silence, either fully or partially. Officers estimated that on average, 59.25% of suspects exercise their right to legal assistance before the police interview (SD = 23.59, N = 130). Of these suspects, 78% were estimated to be guilty (SD = 14.66, N = 113). Respondents further estimated that on average 59.13% of suspects make use of the right to legal assistance during police interviews (SD = 25.23, N = 130), with 78.51% perceived as guilty (SD = 14.77, N = 113). Regarding the right to silence, respondents estimated that 13.88% of suspects in police interviews fully use their right to silence (SD = 16.91; N = 100), out of which 90.87% were believed to be guilty (SD = 13.01; N = 100). Respondents estimated that on average 30.71% of suspects partially use their right to silence (SD = 26.91), of which they believed 81.19% to be guilty (SD = 18.71; N = 108).

A one-way repeated measures analysis of variance (ANOVA) with Bonferroni post hoc comparisons was performed to evaluate whether respondents’ guilt perception of suspects differed between the four situations of using suspects’ rights (use of legal assistance before the interview; use of legal assistance during the interview; partial use of the right to silence; full use of the right to silence). There was a significant situation effect on officers’ guilt perceptions (Wilk’s Lambda = 0.401, F(3,86) = 42.87, p < 0.001, η2 = 0.60). Suspects were significantly more likely to be perceived as guilty if they fully used their right to silence (M = 91.53, SD = 12.64) than if they only partially used their right to silence (M = 81.43, SD = 17.32). Furthermore, suspects were significantly more likely to be perceived as guilty by respondents if they fully used their right to silence (M = 91.53, SD = 12.64) compared to if they used their right to legal assistance before (M = 76.21, SD = 15.55) or during the police interview (M = 76.61, SD = 15.90).

Length, Frequency, and Aim of Suspect Interviews

Officers estimated the average length of a suspect interview to be a little over 1.5 h (SD = 49.25, N = 127). Responses indicating average interview lengths of less than 10 min were not included. The longest suspect interview (including breaks) was estimated at an average of approximately 5 h (SD = 178.93, N = 127). On average, suspects were interviewed 2.4 times within an investigation (SD = 1.13). Regarding the goal of suspect interviews, 86 (66.2%) police officers stated that the suspect interview was aimed at obtaining the truth, 42 (32.3%) believed it aimed at gathering accurate information, and two (1.5%) said that the goal was to obtain a confession.

Beliefs About Deception Detection

On average, respondents estimated their accuracy in judging if a suspect is lying or telling the truth at 71.37% (SD = 13.77). Additionally, respondents were asked if there was a difference in this accuracy level depending on whether the suspect was denying or confessing. The findings showed that most respondents (82.3%; N = 131) believed that it does not make a difference, while 13% said they are better at judging whether a denying suspect is telling the truth and 9% stated that they are better at judging whether a confessing suspect is telling the truth.

(False) Confession Rates

Asked to consider all suspect interviews they had ever been involved in, respondents estimated that on average, 58.74% of suspects provided a partial or full confession (SD = 19.47); 41.53% of interviewed suspects did not admit or confess anything (SD = 19.47). Furthermore, 92 respondents (70.8%) reported having experienced a false confession. Of these officers, the number of estimated experienced false confessions averaged at 21.54 (SD = 66.22, N = 92). However, these frequency observations displayed a significant degree of skewness (5.21, SE = 0.25) and kurtosis (30.27, SE = 0.49). After normalizing the distribution by outlier omission, the estimated number of experienced false confessions averaged at 7.66 (SD = 10.34, N = 87). Of the total number of respondents who experienced false confessions, 87 were reported to be voluntary, while 11 were coerced.

Interviewing Techniques

To examine the use and frequency of various suspect interviewing techniques, investigators were asked to rate the extent to which they use each of 27 different techniques on a 5-point Likert scale (1 = never; 2 = rarely; 3 = sometimes; 4 = often; and 5 = always). Results are presented in Table 2, with techniques ordered in descending order of estimated usage frequency.

Table 2 Self-reported frequency of usage of 27 interviewing techniques
Full size table
Fig. 1
figure 1

Dendrogram representing the cluster of interview techniques

Full size image

A multivariate linear regression model was used to examine whether respondent characteristics could be used to predict the use of each cluster of techniques. Based on Kassin and colleagues’ (2007) study, five predictor variables were tested: “(a) the number of years the respondent had served in law enforcement (experience), (b) whether or not the respondent had attended special training (training), (c) the respondents’ confidence in his or her own lie detection ability (confidence), (d) the number of interrogations conducted by the respondent (interrogations), and (e) the average length of interrogations conducted by the respondent (length)” (p. 391). A regression model was fit for each of the four clusters and results show that only one of the four analyses was significant, F(5,120) = 4.34, p < 0.001, adjusted R2 = 0.153. The more experience the participants had, the more likely were they to use Cluster 3 tactics. None of the other variables were significant in Cluster 3.

A cluster analysis was conducted to identify subgroups of interview techniques that are frequently used together. Manhattan metrics were used to measure the distance between two techniques, indicating that strategically disclosing evidence and confronting the suspect with actual evidence of his/her guilt were most often used conjointly while treating the suspect with respect and physically intimidating the suspect were furthest apart from each other. A hierarchical cluster analysis using the Ward method was conducted to link clusters. Figure 1 shows a dendrogram presenting the model in which the number of clusters was set to four.

Study 2

Method

Participants

A total of N = 208 police investigators from the French-speaking (94.23%, N = 196) and German-speakingFootnote 5 (5.77%, N = 12) parts of Belgium consented to participate in the study. However, 91 officers answered fewer than 24 interviewing technique questions and were excluded from the final sample. Another 64 responses were removed.Footnote 6 Data were screened for non-normality,Footnote 7 leading to the omission of an additional six responses. The final sample contained 47 (89.4% French, 10.6% German) police officers.

Materials

The questionnaire from Study 1 was reused without any content changes. Translation of the original English questionnaire to French and German for the respective subsamples was carried out following the same procedure (Pan and De La Puente 2005) as in Study 1.

Procedure

As in Study 1, only police officers who had participated in a training course on suspect interviewing in the past 10 years were contacted and invited to participate, yet this time only police officers from the French- and German-speaking parts of Belgium were contacted. They received an email with an invitation to participate, background information on the study and its anonymous nature, and a hyperlink to the questionnaire and informed consent form.

Results

Nearly all police officers (97.8%, N = 46) indicated working for the local as opposed to the federal police. The officers in the final sample had a mean age of 41.06 (SD = 7.29) and most of them was male (63%, N = 46). Respondents reported having an average of 17.85 years of experience working in law enforcement (SD = 7.36). As in Study 1, most respondents worked as general criminal investigators (see Table 3 for results).

Respondents reported an average of 10.36 years of experience insuspect interviewing (SD = 5.54) and claimed to have conducted,on average, 35.43 suspect interviews throughout the past year(SD = 25.38). They rated their own interviewing skills with a meanscore of 7.33 (N = 46, SD = 0.73). Most officers (91.3%, N = 46) indicated to have had training in suspect interviewing.

Table 3 Distribution of departments represented in the sample
Full size table

Electronic Recording

Only one officer (2.1%) reported that they always electronically record suspect interviews, while 15 (31.9%) claimed to record interviews only in specific circumstances and 31 (66.6%) did not record interviews at all. Of the 16 respondents who reported using electronic recording, 15 preferred video (93.8%), and one indicated to sometimes use audio and sometimes video (6.3%). Respondents were also asked whether they believed suspect interviews should be recorded. A preference for video recording was indicated by 30 officers (63.8%), three officers (6.4%) preferred audio, and 14 (29.8%) disagreed with mandatory recording altogether.

Suspects’ Rights

As in Study 1, respondents estimated the extent to which suspects use their legal rights and what percentage of suspects who use these rights are, in their opinion, guilty. The officers estimated that on average, 55.57% of suspects use their right to legal assistance before the interview (SD = 24.55), and the majority of suspects who use this right were perceived as guilty (M = 77.45, SD = 16.35). Respondents further estimated that an average of 47.66% (SD = 27.93) of suspects make use of the right to legal assistance during the interview and perceived around 78.51% (SD = 16.45) of these suspects as guilty. Regarding the right to silence, the police officers reported that only a very small percentage of suspects use this right fully (M = 6.66%, SD = 10.11). Furthermore, most suspects who use this right fully were perceived as guilty (M = 91.77%, SD = 13.58). An average of 27.13% of suspects was believed to use their right to silence partially (SD = 21.92), and 84.81% of these suspects were considered guilty (SD = 14.94).

A one-way repeated measures ANOVA with Bonferroni post hoc comparisons revealed a significant effect of suspects’ rights on respondents’ guilt perception of suspects (Wilk’s Lambda = 0.597, F(3,43) = 9.903, p < 0.001, η2 = 0.403). Suspects who fully use their right to silence (M = 91.77%, SD = 13.57) were significantly (p = 0.030) more likely to be perceived as guilty than suspects who only partially use their right to silence (M = 84.81%, SD = 14.94). Furthermore, suspects who use their right to silence fully were significantly (p < 0.001) more likely to be perceived as guilty than suspects who use their right to legal assistance, either before (M = 77.45, SD = 16.35) or during (M = 78.51, SD = 16.45) the interview. Finally, the guilt ratings of suspects who partially use their right to silence were higher (p = 0.043) than those of suspects who use their right to legal assistance before the interview.

Length, Frequency, and Aim of Suspect Interviews

The average duration of suspect interviews was estimated at 125 min (SD = 58.22). Officers estimated the longest suspect interview they ever conducted (including breaks) at approximately 5.5 h on average (SD = 137.22). Suspects were interviewed around 2.74 times (SD.28). Regarding the goal of suspect interviews, none of the police officers reported obtaining a confession to be the main aim. Thirty-one officers (66%) said the goal of a suspect interview is to gather accurate information, while the remaining 16 officers (34%) believed the aim is to obtain the truth.

Beliefs About Deception Detection

The police officers estimated their ability to detect deception in suspects’ statements at 71.38% accuracy on average (SD = 11.83). Most police officers (78.7%) believed the accuracy of their judgments did not depend on whether the suspect was denying or confessing. Six officers (12.8%) believed to be better at deception detection when a suspect is confessing, and four (8.5%) believed their judgments to be more accurate when a suspect is denying.

(False) Confession Rates

Respondents reported that, on average, approximately half of the suspects they had ever interviewed made a partial or full confession (M = 52.34%, SD = 21.16). 46.8% of the officers indicated that at least one of the confessions they had ever obtained was false. These officers were subsequently asked to specify the number of false confessions they had witnessed and on average reported 3.64 false confessions (SD = 2.87). When asked to classify the reported false confessions, 19 (85%) of the officers (N = 20) reported having experienced a voluntary false confession, and 2 (10%) had experienced a coerced false confession. One officer (5%) reported having experienced both voluntary and coerced false confessions.

Interviewing Techniques

As in Study 1, police officers reported how often they use several interviewing techniques on a 5-point Likert scale (1 = never; 2 = rarely; 3 = sometimes; 4 = often; and 5 = always). Results are presented in Table 4, with techniques ordered in descending order of estimated usage frequency.

Table 4 Self-reported frequency of usage of 27 interviewing techniques
Full size table

Discussion

The present research findings provide insight into self-reported practices and beliefs of both Flemish police officers and officers from the French- and German-speaking regions of Belgium. Data were gathered to compare to each other and to observations from other countries as well as to prior research in Belgium. Due to the low number of federal police officers participating, there is no comparison possible between local and federal police.

Electronic Recording

The Belgian police officers generally showed a favorable attitude toward standardizing electronic recording of all suspect interviews even though the reported rate at which interviews are actually recorded was very low. This contrast aligns with prior research (Kassin et al. 2007; Schell-Leugers et al. 2023; Vanderhallen et al. 2022). Given the predominantly supportive attitude, it seems unlikely that the low number of recorded interviews in Belgium is due to police officers’ reluctance to do so. There is likely a lack of recording equipment available to police departments (Vanderhallen et al. 2017). Making electronic recordings of suspect interviews is not mandatory in Belgium (Art. 112ter, Sv.).

Length, Frequency, and Aim of Suspect Interviews

In alignment with Leo’s (1996, 2008) findings and recommendations, suspect interviews in Belgium do not appear to exceed two hours in duration. The reported mean durations in the present research (1.5–2 h) also align with average interview lengths reported in other studies (Areh et al. 2016; Kassin et al. 2007). The mean duration of the reported longest interview was slightly higher according to Belgian police officers (± 5–5.5 h) than reported by police officers from the US (4.21 h; Kassin et al. 2007) and Slovenia (4 h; Areh et al. 2016). However, the average duration of the longest interview reported in the present study is below the six-hour threshold posed by Feld (2006) and Blair (2005) for coercive interviews. Thus, although it cannot be guaranteed that longer interviews do not occur, the interview durations reported by Belgian police officers are within acceptable boundaries.

Respondents in the present study reported that each suspect was interviewed two to three times on average, which is very similar to findings from previous studies. For instance, Kassin et al. (2007) reported that in the US, suspects are interviewed around three times. The Belgian officers’ estimate is also similar to that of Slovenian (2; Areh et al. 2016), Spanish (2.22; Schell-Leugers et al. 2023), and Dutch police officers (2.89; Vanderhallen et al. 2022).

The current research suggests an improvement in Belgian police officers’ beliefs regarding the goal of suspect interviews since Ponsaers and colleagues’ (2001) study, which found that only 21.9% believed gathering verifiable information was the main aim. In Study 1, only 1.5% considered obtaining a confession the aim, dropping to 0% in Study 2, with the majority believing in gathering verifiable information or obtaining the truth. This indicates a shift towards objective investigative interviews in Belgium (Soukara et al. 2009), though conducting interviews to “obtain the truth” is still not ideal. Police may believe they know the truth, influencing their evaluation of what constitutes truth (Horselenberg and Van Koppen 2017), thus requiring nuance in interpreting the observed positive shift.

Suspects’ Legal Rights

Police officers’ beliefs and estimations about suspects’ use of their right to legal assistance differed from prior research findings. For instance, Penne and colleagues (2013) found that after the introduction of the Salduz Act in 2012, approximately 72% of suspects in Belgium used their right to legal assistance before and during the interview. Officers in the present research estimated this number to be lower (± 55–59%). This difference may result from a change in the Salduz Act in 2016. Before 2016, suspects could only waive these rights after a mandatory confidential phone consultation with a lawyer, which, since the implementation of Salduz Bis, is no longer necessary. Thus, it became easier for suspects to waive their rights, possibly explaining the lower frequencies. Indeed, more recent data from the Flemish Bar Association support our findings. Of the 41,449 first suspect interviews conducted in Belgium in 2018, 61% asked for legal assistance, 1.6% asked for a confidential phone consultation, and 37.4% waived their right to legal assistance.

The present findings on suspects’ use of their right to silence appear to be consistent with earlier studies. According to our respondents, more Belgian suspects appear to use their right to silence partially (27.13–30.71%) than fully (6.66–13.88%), which has also been found for suspect interviews in the Netherlands (34.4% full, 49.1% partial; Vanderhallen et al. 2022), but not in Spain (79.1–82.9% full, 25.4–30.7% partial; Schell-Leugers et al. 2023). Compared to Dutch suspects (Vanderhallen et al. 2022), a far smaller percentage of Belgian suspects appears to make use of their right to silence, whether fully or partially. The observation that very few suspects in Belgium use their right to silence fully is consistent with prior research in Belgium (± 3–16%; Beazley et al. 2021a, b; Vanderhallen et al. 2014) that has been conducted since Penne et al. (2013).

Regarding perceptions of guilt, prior research (e.g. Beazley et al. 2021b; Kassin et al. 2007) has shown that police tend to associate the waiver of rights with innocence and the use of rights with guilt. The present research confirms this finding as most suspects (± 78–92%) who use their rights were perceived as guilty. Especially the use of the right to silence (either fully or partially) was implicitly believed to be indicative of guilt. Thus, when a suspect makes use of their rights, Belgian police officers appear more likely to perceive them as guilty. This finding aligns with recent findings from Spain (Schell-Leugers et al. 2023) and the Netherlands (Vanderhallen et al. 2022).

Deception Detection

Like Kassin et al. (2007), Belgian police officers overestimated their own deception detection abilities at approximately 71% in both studies. The observed overestimation is similar to Dutch police investigators’ self-perception of deception detection skills (67.9%; Vanderhallen et al. 2022), but slightly lower than observed among Spanish police investigators (± 80.8%; Schell-Leugers et al. 2023). Interviewers’ overestimation of their lie detection ability can bias them into using techniques aimed at eliciting a (potentially false) confession, especially when they believe the suspect is guilty (Kassin and Gudjonsson 2004; Meissner and Kassin 2002). Belgian police officers’ self-perceptions regarding deception detection ability were largely supported by prior research showing that people’s lie detection accuracy remains unchanged whether suspects are denying (Meissner and Kassin 2002) or confessing (Kassin et al. 2005).

(False) Confession Rates

Belgian police officers estimated that the number of interviewed suspects that confess either partially or entirely is between 52 (Study 2) and 59 (Study 1) percent. These numbers are lower than estimates from the UK (Clarke and Milne 2001; Moston et al. 1992; Pearse and Gudjonsson 1996; Phillips and Brown, 1998) and the US (Kassin et al. 2007; Leo 1996), but similar to recent estimates from the Netherlands (Vanderhallen et al. 2022). Between 48% (Study 2) and 71% (Study 1) of the police officers reported having experienced at least one false confession. They further estimated to have encountered, on average, between four (Study 2) and eight (Study 1) false confessions since starting to conduct suspect interviews. Respondents also reported a few coerced false confessions, which could indicate that coercive, confession-focused interviewing techniques are still being used. Overall, the false confession rates observed in the present research are surprising, especially when considering our other findings (e.g. regarding the goal, duration, and frequency of suspect interviews) are in line with the guidelines for minimizing chances of eliciting false confessions. Given we asked about false confessions throughout the entire career of the respondents, it could be that most of the false confessions happened before the shift from accusatorial to information-gathering techniques.Footnote 8

Interview Techniques

Flemish police officers appear to still use rapport building as a technique less frequently than police officers in other countries. Even though the use of this technique increased from 20.2% in at least 25% of cases (Ponsaers et al. 2001) to at least 78% using it sometimes or more often, US officers reported using this technique often rather than sometimes (Kassin et al. 2007). In the US, 32% claimed to always establish rapport with the suspect (Kassin et al. 2007), compared to only 17% of the Flemish respondents in this study. The Flemish findings on rapport building did not only differ markedly from previous US findings but also from the French- and German-speaking Belgian police officers. In Study 2, almost 98% of respondents reported using this technique at least sometimes, which is more comparable to the US findings.

When it comes to how suspects are treated, Belgian police officers appear to have improved their attitude significantly since 2001. While Ponsaers and colleagues (2001) found that only 22.6% of respondents indicated they treat the suspect with respect in at least 25% of cases, all officers in the current research indicated using this technique at least often. The results further suggest that frustration and anger are nowadays only expressed at the suspect rarely, indicating a shift towards a civil, respectful relationship with the suspect. While physical intimidation was used by 1.1% in more than 25% of cases in 2001 (Ponsaers et al. 2001), no respondent indicated they use this tactic in the current study. The Belgian investigators also appear to respect suspects’ right to legal assistance as they reported rarely to never discouraging interventions by the suspect’s lawyer, and approximately 85–91% of respondents said they never confront the suspect with false evidence. Finally, the use of minimization, maximization, and moral justification techniques has clearly decreased since Ponsaers and colleagues’ (2001) investigation. However, they are still used sometimes which confirms the findings by Beazley and colleagues (2021b).

Unfortunately, the usage frequency of some accusatorial techniques seems to have increased rather than decreased among Belgian police officers since 2001. For example, the Belgian officers reported confronting suspects with photographs of the victim more frequently (± 31–40% at least sometimes) than in 2001 (6.2% in more than 25% of cases; Ponsaers et al. 2001). This might seem surprising considering that under accusatorial models, victim photos are often used to move the suspect toward confessing (Meissner et al. 2014), thus indicating an unfavorable shift back to accusatorial interviewing styles among Belgian police officers. Alternatively, the technique might nowadays be used to complete an already-made confession rather than elicit one. Furthermore, Belgian police officers appear to more frequently pretend to have evidence against the suspect (sometimes or more by about 50% of respondents) than in 2001 (12% in at least 25% of cases; Ponsaers et al. 2001). In our questionnaire, “hypothetical questions” were added in brackets to the item “pretending that there is evidence against the suspect”. After the shift towards a more information-gathering approach, pretending to have evidence is no longer accepted. Instead, asking hypothetical questions in which for example results from fingerprints are suggested as possibilities rather than as findings, came more into place since they are being taught in some interview courses (Vanderhallen et al. 2016).

Overall, the analysis of the techniques showed a grosso modo shift toward a more information-gathering style compared to findings from 2001. This may partially be unsurprising as Ponsaers and colleagues (2001) also presented more accusatorial techniques for review and Beazley and colleagues (2021a, b) only focused on interviewing practice in relation to suspect rights specifically. Nonetheless, it is still a favorable outcome to minimize the risk of miscarriages of justice. There is a clear tendency to use more information-gathering techniques, but accusatorial techniques and deception detection techniques still seem to be in place.

It is important to note, however, that further improvements in interviewing practice in Belgium arguably require police officers to follow particular training in the use of information-gathering interviewing techniques. However, suspect interviewing training is insufficiently and not always consistently offered across Belgian police training institutes (Vanderhallen et al. 2016). The Basisverhoortechniek that serves as the basic interviewing training in Belgium (Van de Plas 2007) only provides basic knowledge on interviewing in general, which is taught by local police academies. The additional obligatory functional training for detectives that is taught at the National Criminal Investigation Academy features training on interviewing victims, witnesses, and suspects, and other advanced training courses are offered by both the national academy and private organizations. Despite efforts to standardize all interview training in the past decennium, there are still differences regarding content and approach in these programs because the training is delivered at various academies and by various instructors (Vanderhallen et al. 2016). For example, interview training handbooks in Belgium recommend diverse techniques, with some handbooks recommending questionable procedures (Bockstaele 2018a, b) and others exclusively supporting information-gathering methods (Schellingen and Scholten 2014). Overall, Belgian police education does not completely abandon more accusatorial and controversial techniques, which are criticized in the literature as ambiguous and enhancing the risk for, for example, false confessions (Vanderhallen et al. 2016).

Limitations

Notwithstanding the relevance of its findings, several limitations impact the generalizability of the present research. First, the study suffered from notable sampling issues. No random sampling was used and not all police departments were reached, and participation was voluntary. Furthermore, most respondents worked for the local police, leaving federal police officers largely under-represented. Additionally, the sample from Study 2 was small. Ideally, future research should seek to better represent federal police officers as well as obtain sufficient and similar sample sizes for all three Flemish, French-speaking, and German-speaking Belgian police investigators to allow for the possibility of within-country comparisons.

Second, the collected data concern perceptions, and only in exceptional circumstances are details of every case remembered. As respondents estimated different frequencies and numbers based on their experiences and beliefs, they likely based their answers on estimation and inference strategies which do not necessarily accurately represent the state of events in question (Schwarz, 2007). Questions concerning an attitude (e.g. whether suspect interviews should be recorded) lead respondents to either answer corresponding to a previously formed attitude from their memory or to spontaneously form a judgment depending on the information available at that moment. Such inference strategies can have a biasing effect (Schwarz et al. 1998; Tourangeau et al. 2000).

Third, despite conducting pretests to limit this problem, some participants remarked that some questions were still unclear or could be interpreted in different ways. For example, the question about the crime type respondents mostly dealt with was excluded from the analysis because participants often mentioned their department instead of the predominant crime type. Additionally, some respondents found it challenging to answer questions about the guilt or innocence of suspects who exercised their rights, citing difficulties in estimating guilt, uncertainty inherent in determining guilt, or the inability to express doubt (which is common in practice). Lastly, respondents also addressed the fact that their answers could differ depending on the type of crime they interviewed suspects about and whether the suspect was caught in the act or not. Taken together, the questionnaire used did not allow for sufficient nuance and differentiation and calls for a more qualitative follow-up study.

Conclusion

As part of a Europe-wide project, the present research aimed to gain insight into Belgian suspect interviews by examining police officers’ self-reported practices, perceptions, and beliefs about various issues related to suspect interviews. Our findings regarding the duration, frequency, and recording of suspect interviews are in line with previous research and follow the guidelines to minimize the chances of false confessions. In addition, Belgian police officers for the most part appear to rely on interviewing techniques that fit within the information-gathering approach. However, officers still reported using some accusatorial tactics and a higher number of false confessions than in other countries. These findings suggest that the interviewing training at Belgian police academies could benefit from more uniformity and a clear focus on information-gathering interviewing techniques only. In addition, standardizing electronically recording suspect interviews would greatly facilitate future studies in more objectively examining what happens in the interviewing room.